Klook K.ai: 95% Faster Updates, Operators Want Sales Help Klook launched K.ai, an AI agent built into its merchant portal, on November 3, 2026, and says early users completed individual inventory and pricing updates 95% faster and batch checks 97% faster. K.ai retrieves business information, answers booking and availability questions, updates inventory and pricing, generates booking reports, and executes merchant instructions, plus a one-click product listing feature that pre-fills structured listings for review. Klook has not disclosed K.ai's price or the number of merchants using it, and operators using the agent say their next priority is help with selling, pricing and market guidance rather than back-office admin. Klook gave its experiences merchants an AI agent on November 3, 2026, and the first operators using it say it clears their back-office chores fast. What they want next is help selling. That tension runs through Skift’s report on K.ai https://skift.com/2026/11/17/klooks-ai-agent-handles-the-admin-but-operators-want-help-selling/ , which pairs Klook’s speed claims with a veteran tour operator’s wish list for pricing and market guidance. K.ai sits inside Klook’s merchant portal. It retrieves business information, answers booking and availability questions, updates inventory and pricing, generates booking reports, and carries out instructions on the merchant’s behalf. Klook says early users finished individual inventory and pricing updates 95% faster and batch checks 97% faster. Agents that act inside business software, rather than just chat about it, raise fresh questions about control, which we covered in our piece on who secures AI agent conversations https://shattered.io/securing-ai-agent-to-agent-communication/ . This analysis separates what Klook and Skift have confirmed from what remains open. It covers the product, the numbers, the one operator voice on record, the competitive logic, the risks, and five predictions for how merchant-side agents develop through 2027. Where the available reporting runs out, the article says so plainly. Don't miss new tech stories on Google Add Shattered once in the Google app and our stories show up in your news suggestions. Add now https://www.google.com/preferences/source?hl=en&q=https%3A%2F%2Fshattered.io%2F What Klook Launched on November 3 The product is called K.ai, and its job is narrow and practical. Klook built it into the merchant portal that tour operators, activity providers and attraction sellers already use to run their listings. Merchants do not install anything new or open a separate app. They type or speak an instruction, and the agent acts on it inside the tools they already know. The launch date matters for context. Klook rolled K.ai out on November 3, 2026, and the company is deploying it in phases around the world ahead of the year-end travel season, according to The Paypers https://thepaypers.com/payments/news/klook-launches-kai-agent-to-manage-merchants-back-office-operations . Year-end is when many operators change prices, add departures and watch capacity most closely, so the timing puts the agent under real load almost immediately. Klook has confirmed five core capabilities. K.ai retrieves business information, answers booking and availability questions, updates inventory and pricing, generates booking reports, and executes merchant instructions in the portal. A sixth feature, one-click product listing, turns uploaded materials into structured listings and pre-fills fields so the merchant can review and edit before anything goes live. What the sources do not provide is equally important. No price for K.ai or for Klook’s merchant services has been published. No count of merchants using the agent has been disclosed. Any article claiming either figure is guessing, and this one will not. How K.ai Works Inside the Merchant Portal Think of K.ai as a command layer over the portal. A merchant who once clicked through calendars, rate tables and order lists can now state an intent in plain language. The agent translates that intent into portal actions, then shows the result. Rise Productive described the launch as an AI assistant that can change merchants’ inventory https://www.riseproductive.com/news/klook-kai-merchant-portal-agent-20261103 and related settings, which captures the shift from answering questions to taking actions. The weekend price example Klook’s own example shows the pattern. A merchant asks for a report on the previous week’s bookings, reads it, and then tells K.ai to raise weekend prices by 10%. The change lands in the merchant portal. In a hypothetical case, a weekend tour listed at $100 would move to $110 after that single instruction. The merchant never opens a pricing screen. That two-step flow, report first and action second, matters. The merchant reads data and decides, and the agent handles the mechanical part. It keeps the human in the judgment seat while removing the clicking. One-click listing from uploaded materials The listing feature targets a different bottleneck. Creating a new product on a booking marketplace means filling many structured fields, from descriptions and inclusions to schedules and meeting points. K.ai generates a structured listing from materials the merchant uploads, pre-fills the fields, and hands it back for review and editing. The merchant stays the editor of record, and the agent acts as a first-draft writer. The Numbers Klook Reports: 95% and 97% Klook’s headline figures come from early users. Individual inventory and pricing updates ran 95% faster than before, and batch checks across products, inventory and orders ran 97% faster. These are company-reported numbers. Klook has not published the sample size, the baseline task times, or how it measured the comparison, so they work best as directional evidence. The wording leaves room for interpretation. If the 95% figure means a 95% cut in time spent, a task that took 20 minutes now takes about one minute. If it means 95% more throughput, the gain is far smaller. Read the same way, a 97% cut turns a 20-minute batch check into roughly 36 seconds. The arithmetic here is ours, not Klook’s, and it assumes the time-reduction reading. | Item | Figure | Source and status | |---|---|---| | Launch date | November 3, 2026 | Klook announcement, confirmed | | Rollout | Phased, global, before year-end travel season | Klook announcement, confirmed | | Individual inventory and pricing updates | 95% faster | Klook-reported, early users | | Batch checks of products, inventory, orders | 97% faster | Klook-reported, early users | | SME share of experiences sector | About 95% | Klook-reported sector figure | | Example price change | Weekend prices up 10% | Klook’s recorded example | | K.ai price | Not disclosed | Unconfirmed | | Number of merchants using K.ai | Not disclosed | Unconfirmed | The last two rows deserve attention. Speed claims are cheap to publish and expensive to verify. Until Klook shares adoption numbers and methodology, the figures describe what the company says early users experienced, not what the typical merchant will see. Jo Wong: A 35-Person Operator Ten Years on Klook Skift anchors its story on Jo Wong, founder of a tour-guide and day-tour supplier. The business has sold on Klook for more than 10 years and employs 35 staff. That profile fits the operator Klook says it built K.ai for: established, busy, and short on administrative hours. Wong told Skift the business uses K.ai to review bookings, spot products approaching capacity, and compile complex data faster. In Klook’s announcement, Wong put the experience this way: “Now I tell K.ai what I want, and it’s done before my coffee is.” It is a vivid line, and it describes the admin win exactly. The agent turns a chore into a sentence. The wish list is the more interesting part. Wong wants pricing recommendations, market-price comparisons and booking-trend analysis. None of those are described in the available reports as current K.ai features. They are requests, and they point at the next product decision Klook faces. One operator is not a survey. Skift documents Wong’s requests clearly, but the reporting does not establish that operators in general want the same thing. We treat Wong as a credible signal from a long-tenured seller, not as proof of a universal demand. Coverage by The World of AI https://theworldofai.org/ai-news/2e550e14/ confirms the launch itself but adds no merchant survey either. The Gap Between Admin Automation and Selling Help The story’s central point is a gap in the product. Admin automation answers how to finish a task faster. Selling help answers a harder question, which is what to do to earn more. The first needs access to the portal and a reliable way to execute instructions. The second needs judgment built on market data, competitor prices, seasonality and demand signals. K.ai, as described, handles the first well. It cannot say whether a 10% weekend increase is wise. It applies the increase once the merchant decides. Wong’s three asks pricing recommendations, market-price comparisons, booking-trend analysis all push the agent from executor toward advisor. That move is not trivial. An advisor agent needs data the merchant portal may not hold, such as what rival operators charge for similar tours on a given Saturday. It also takes on liability. If a recommendation drives a price too high and bookings drop, the merchant will blame the tool. An executor agent carries far less of that risk, because the human supplied the decision. | Capability | In K.ai reports? | Operator request Skift | |---|---|---| | Retrieve business information | Yes | Used to review bookings | | Answer booking and availability questions | Yes | Used to find products nearing capacity | | Update inventory and pricing | Yes | Not a stated gap | | Generate booking reports | Yes | Used to compile complex data faster | | One-click product listing | Yes, with merchant review | Not a stated gap | | Pricing recommendations | Not described | Requested | | Market-price comparisons | Not described | Requested | | Booking-trend analysis | Not described | Requested | The pattern in the table is clean. Everything operational is covered. Everything strategic is requested. Booking reports sit closest to the boundary, since a report can describe what happened without saying why or what to do next. Why Small Operators Are the Target Klook says small and medium-sized operators make up approximately 95% of the experiences sector. If that holds, the typical merchant on the platform is a small team with no revenue analyst, no pricing department and no spare engineer. Wong’s 35 staff already put the business toward the larger end of that group. For such a merchant, the scarce resource is attention. A guide who spends mornings adjusting availability is not planning routes or training staff. Klook co-founder and CTO Bernie Xiong framed the intent in comments reported by FutureCIO, saying K.ai brings “the most practical and valuable agentic AI capabilities into their day-to-day operations, removing much of the manual work involved in running their businesses, so they can spend more time on what they do best: delivering meaningful experiences for travellers.” The business logic is plain. Klook earns when bookings happen on its marketplace. Merchants who keep inventory current and prices competitive sell more, and merchants buried in admin let calendars go stale. An agent that keeps listings fresh helps the supplier and the platform at once. The small-operator skew also explains why the admin pitch comes first. Time savings are easy to demonstrate and easy to trust. Revenue advice from an algorithm is harder to sell to a business owner who has run tours for a decade and knows the local market by instinct. Expert and Operator Voices Three on-record statements frame how Klook and its merchants describe K.ai. We quote them verbatim and add nothing beyond the attributions in Klook’s materials and the reporting above. Jo Wong , founder of a tour-guide and day-tour supplier with 35 staff, said in Klook’s announcement: “Now I tell K.ai what I want, and it’s done before my coffee is.” Wong speaks for the speed benefit, and Skift separately records the requests for pricing and market guidance. Erica Yu , co-founder of MyProGuide, described the work K.ai is meant to replace in Klook’s newsroom post: “I used to block out an hour every morning to update prices and availability across my tours.” If that hour repeats every weekday, it adds up to about five hours a week, which shows why a 95% claim, if it holds, would matter to a small team. Bernie Xiong , Klook’s co-founder and CTO, gave the company’s rationale, quoted in the previous section. His emphasis falls on removing manual work so merchants can focus on delivering experiences. The statement leaves something out. It says nothing about helping merchants sell more, which is the gap Skift’s headline identifies. Two of three voices are merchants, and both describe time savings. The only mention of sales guidance comes from Wong’s wish list. That mix of evidence supports a measured conclusion: the admin benefit is documented, and the selling benefit is an open request. Competitive Context: Where Merchant-Side Agents Fit Klook is not alone in pushing agents toward the supply side of travel and commerce. The broader pattern is that platforms first built AI for shoppers, with chat planners and recommendation tools, and are now turning to the sellers who fill their catalogs. Klook’s own newsroom lists a separate announcement about partnering with Meta’s Muse on agentic travel experience booking in Asia https://www.klook.com/newsroom/klook-meta-muse-agentic-travel-booking/ , which shows the company working on both ends of the transaction. Seller-side agents compete on three axes. The first is action coverage, meaning how many portal tasks the agent can complete without a human click. The second is decision support, meaning whether it advises or merely obeys. The third is trust, meaning how safely it handles changes that touch money. K.ai scores well on the first, appears light on the second, and has not published enough about the third to judge. Other vendors approach the problem from different angles. Our earlier reporting on Perplexity’s agent teammates for ongoing work https://shattered.io/perplexity-teammates-ai-agents-ongoing-work-200-max-2026/ showed general-purpose agents moving into persistent business tasks, and our coverage of Firmly’s $15M Series A for agentic commerce https://shattered.io/firmly-15m-series-a-squarepoint-agentic-commerce-ask-agent-2026/ showed investors backing agents on the selling side. A marketplace-native agent like K.ai has one clear edge over both: it already holds the merchant’s inventory, prices and booking history. That edge cuts two ways. Data access makes advice possible, since Klook can see demand across its own listings. It also creates a conflict question that Klook will have to answer. A marketplace that recommends prices to its own suppliers sits close to the line between helping them and steering them. Historical Context: From Phone Calls to Agents Tour and activity operators have absorbed several waves of tooling. Bookings once arrived by phone, email and walk-in. Online marketplaces then pulled demand onto platforms, and merchants gained extranet portals to manage listings. Channel-management software followed for operators selling across several platforms, and each step cut some manual work while adding a new dashboard to learn. Agents represent a different kind of change. Earlier tools asked the merchant to learn the software’s vocabulary. An agent learns the merchant’s vocabulary instead. The shift from forms to sentences is the real story behind K.ai, and it explains why the product pitch centers on telling the agent what you want. The same arc has played out in other software categories. Assistants first answered questions, then drafted content, and now they act inside systems of record. Each step moves the risk profile. Answering wrongly wastes a minute, while changing a price wrongly can cost a weekend of bookings. The merchant portal is a clean test of that jump, because the actions are bounded and easy to check. Falling model costs also help explain the timing. Vendors keep claiming cheaper inference, as in OpenAI’s GPT-6.2 Astra claim of a one-third cost https://shattered.io/openai-gpt-6-2-astra-5-2-point-gain-one-third-cost/ , and cheaper reasoning makes it viable to put an agent in front of every small merchant. Klook has not said which model powers K.ai, so we draw no link between the two. We note only that the economics of serving millions of small instructions have improved across the industry. Risks When an Agent Changes Your Prices Letting software edit live prices and inventory deserves scrutiny. A mistyped instruction, a misread request or a malicious input could publish a bad rate or close a sold-out tour. The reports we reviewed do not detail approval steps, undo options, audit trails or spending limits for K.ai. We are not suggesting those protections are missing. We are saying the public record does not show them. Security researchers worry about a specific class of attack when agents read content they did not write. A listing description, a customer message or an uploaded file could carry hidden instructions. Our walkthrough of defending against indirect prompt injection https://shattered.io/indirect-prompt-injection-defense-taint-gate-12-steps-2026/ explains why agents that both read untrusted text and take actions need a gate between the two. K.ai’s one-click listing feature reads merchant-uploaded materials, so the question applies directly, though Klook has not commented on it in the sources we found. Cost control is a second concern. Agents that run many steps per instruction can run up usage bills, a problem our report on Forrester’s three agent limits and AI bill shock https://shattered.io/ai-bill-shock-forrester-three-agent-limits-openai-invoice-2026/ examined for enterprise buyers. With no price published for K.ai, merchants cannot yet judge whether usage will be bundled into existing fees or metered separately. A practical safeguard for any operator is simple. Keep a record of prices and inventory before large instructions, check the portal after batch changes, and avoid sweeping commands during peak booking periods until the agent has earned trust on small ones. Market Impact: What It Means for Klook and Its Rivals For Klook, K.ai is a retention and quality play. Merchants who save hours each week have less reason to drift to rival platforms, and listings that stay current convert better. The strategic upside depends on one question: whether the agent can graduate from clerk to advisor. Rival booking platforms face pressure to match. Merchant portals have long been a quiet competitive front, because suppliers often list on several marketplaces and favor the one that costs them least effort. An agent that halves the work of maintaining a listing raises the bar for everyone. We expect competitors to announce comparable tools, though nothing in the sources confirms any specific rival launch. For merchants, the impact is uneven. Operators with the most listings gain the most from batch checks, and small teams gain from reclaimed mornings. Operators who treat the portal as a monthly chore may see less. Merchants who master the agent early will update prices more often and respond faster to demand shifts, which can open a gap over slower neighbors. For travelers, the effect is indirect but real. Fresher availability means fewer cancellations caused by stale calendars. More frequent price changes could also mean more dynamic pricing in tours, a trend consumers sometimes dislike. The weekend 10% example hints at that pattern. What Remains Unconfirmed Good reporting marks its edges. Several points sit outside the verified record, and readers should treat claims about them with caution. - Universal operator demand. Skift documents the requests Jo Wong made for sales help. It does not establish that operators broadly share them. - Pricing. No price for K.ai or Klook’s merchant services appears in the available sources. - Adoption. Klook has not disclosed how many merchants use K.ai. - Methodology. The 95% and 97% figures come from early users, with no published sample size or baseline. - Safeguards. The reports do not describe approval flows, undo features or audit logs for agent-made changes. None of these gaps sink the story. They simply define what the next round of reporting should chase. Klook can close most of them with a single detailed product post. Five Predictions for Merchant-Side Agents These are our analytical forecasts, not reported facts, and none comes from Klook or Skift. - Klook adds advisory features within a year. The wish list from Wong maps to features a marketplace can build from data it already holds. We expect pricing suggestions or demand insights to appear first, probably labeled as recommendations the merchant must approve. - Rivals ship their own portal agents. Once one major platform proves merchants like the experience, competing marketplaces will follow. Expect announcements framed around hours saved. - Merchants demand audit trails. As agents touch live prices, operators will ask for change logs and one-click reversals. Platforms that offer them early will earn trust faster. - Speed claims face scrutiny. Independent operators and trade press will test the 95% and 97% figures against real workloads. The numbers may hold for simple tasks and shrink for messy ones. - The sales-help gap becomes the main battleground. Admin automation will turn into a baseline expectation. Differentiation will move to which agent gives the best pricing and trend guidance. For broader context on where agents and models are heading, see our AI and machine learning coverage https://shattered.io/ai-machine-learning/ . What Operators Should Do Now Operators on Klook have a short checklist. Start with low-stakes tasks, such as pulling a weekly booking report, before handing the agent price changes. Compare the agent’s output against the portal on the first few runs. Treat one-click listings as drafts, since Klook itself positions them for review and editing. Track your own time savings. If Klook’s 95% figure is accurate, a daily updater like the one Erica Yu described should notice within a week. If you do not see the gain, tell Klook, because early feedback shapes what gets built next. Also keep your own market watch going. Until K.ai offers market-price comparisons, rival pricing is still your job. Finally, ask for what you need. Wong’s three requests are public, and a product team can read that kind of signal. Operators who state clearly what decisions they want help with give Klook a roadmap to build against. Frequently Asked Questions What is Klook K.ai? K.ai is an AI agent built into Klook’s merchant portal for experiences merchants. It retrieves business information, answers booking and availability questions, updates inventory and pricing, generates booking reports, and executes merchant instructions in the portal. When did Klook launch K.ai? Klook launched K.ai on November 3, 2026. The company is rolling it out in phases around the world ahead of the year-end travel season. How much faster is K.ai, according to Klook? Klook reports that early users completed individual inventory and pricing updates 95% faster and batch checks of products, inventory and orders 97% faster. The company has not published sample sizes or methodology for these figures. Can K.ai change prices on its own? It changes prices when a merchant instructs it to. In Klook’s recorded example, a merchant reviews last week’s bookings and then tells K.ai to raise weekend prices by 10%, and the change applies in the portal. The reports do not describe the agent setting prices without an instruction. Does K.ai recommend prices or compare market rates? The available reports do not describe it doing so. Jo Wong, whose business has sold on Klook for more than 10 years, told Skift about wanting pricing recommendations, market-price comparisons and booking-trend analysis, which suggests those features are not part of the current offering. How much does K.ai cost? No price has been disclosed. The sources we reviewed give no figure for K.ai or for Klook’s merchant services. How many merchants use K.ai? Klook has not published a user count. Its performance figures come from early users, and the sector figure it cites says small and medium-sized operators make up approximately 95% of the experiences sector. Do operators generally want K.ai to help them sell more? Skift documents that Jo Wong wants this. The reporting does not show that operators across the sector share that view, so treat it as one credible signal rather than a survey result.